Confidence Estimation for Black Box Automatic Speech Recognition Systems Using Lattice Recurrent Neural Networks

Kastanos, Alexandros; Ragni, Anton; Gales, Mark J. F.

doi:10.48550/arxiv.1910.11933

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[20] shows how to optimize black box ASR systems. and [21] shows how to improve confidence estimates produced by such black-box systems. Another closely related work [10] is to use a domain-specific language model and a semantic parser to rescore the hypotheses from a blackbox ASR system.…”

Section: Related Workmentioning

confidence: 99%

Black-Box Adaptation of ASR for Accented Speech

et al. 2020

View full text Add to dashboard Cite

We introduce the problem of adapting a black-box, cloud-based ASR system to speech from a target accent. While leading online ASR services obtain impressive performance on mainstream accents, they perform poorly on sub-populations -we observed that the word error rate (WER) achieved by Google's ASR API on Indian accents is almost twice the WER on US accents. Existing adaptation methods either require access to model parameters or overlay an error correcting module on output transcripts. We highlight the need for correlating outputs with the original speech to fix accent errors. Accordingly, we propose a novel coupling of an open-source accent-tuned local model with the black-box service where the output from the service guides frame-level inference in the local model. Our fine-grained merging algorithm is better at fixing accent errors than existing word-level combination strategies. Experiments on Indian and Australian accents with three leading ASR models as service, show that we achieve upto 28% relative reduction in WER over both the local and service models.

show abstract

Section: Related Workmentioning

confidence: 99%